Using single Conspiracy Number to analyze game progress patterns

Conspiracy Number Search (CNS) is a MIN/MAX tree search algorithm, trying to guarantee the accuracy of the MIN/MAX value of a root node. It suffers from a low efficiency because of its slow convergence and a big cost of computing conspiracy numbers. However, the conspiracy number is still a promising concept for measuring the “stability”, which can be used to analyze game progress patterns. In this paper, we propose a method to analyze game progress patterns in Chinese Chess by using the single conspiracy number. We define the single conspiracy number and apply it in an alpha-beta search framework. Experiments are done on the ElephantEye program which is one of the most powerful AI engines of Chinese Chess. Experimental results show that single conspiracy number can cover the shortage of conventional heuristic values and give more information on the game progress.